COVID-19

Today is

Sys.time()

[1] "2020-09-10 14:37:14 +07"

Loading COVID-19 dataset

library(dplyr)
library(ggplot2)
library(plotly)

set.seed(1234)

filename <- "covid.csv"
#if (!file.exists(filename)) {
  download.file("https://covid.ourworldindata.org/data/owid-covid-data.csv", destfile = filename)
#}
covid_raw <- read.csv("covid.csv", header = T)
covid <- covid_raw

Fix total_cases

covid$date <- as.Date(covid$date)

covid$total_cases_fixed <- sapply(1:nrow(covid), function (n) {
  label <- "total_cases"
  x <- covid[[label]][n]
  
  if (is.na(x)) {
    this_loc <- subset(covid, location == covid$location[n])
    m <- mean(c(
      max(0, subset(this_loc, date < covid$date[n])[[label]], na.rm = T),
      min(max(covid[[label]], na.rm = T), subset(this_loc, date > covid$date[n])[[label]], na.rm = T)
    ))
    
    if (is.na(m)) {
      return(0)
    }

    return(m)
  }
  
  return(x)
})

select(subset(covid, location == "Thailand"), location, date, total_cases, total_cases_fixed)

Fix total_cases_per_million

covid$total_deaths_per_million_fixed <- sapply(1:nrow(covid), function (n) {
  label <- "total_deaths_per_million"
  x <- covid[[label]][n]
  
  if (is.na(x)) {
    this_loc <- subset(covid, location == covid$location[n])
    m <- mean(c(
      max(0, subset(this_loc, date < covid$date[n])[[label]], na.rm = T),
      min(max(covid[[label]], na.rm = T), subset(this_loc, date > covid$date[n])[[label]], na.rm = T)
    ))
    
    if (is.na(m)) {
      return(0)
    }

    return(m)
  }
  
  return(x)
})

covid$total_cases_per_million_fixed <- sapply(1:nrow(covid), function (n) {
  label <- "total_cases_per_million"
  x <- covid[[label]][n]
  
  if (is.na(x)) {
    this_loc <- subset(covid, location == covid$location[n])
    m <- mean(c(
      max(0, subset(this_loc, date < covid$date[n])[[label]], na.rm = T),
      min(max(covid[[label]], na.rm = T), subset(this_loc, date > covid$date[n])[[label]], na.rm = T)
    ))
    
    if (is.na(m)) {
      return(0)
    }

    return(m)
  }
  
  return(x)
})

select(subset(covid, location == "Thailand"), location, date, total_deaths_per_million, total_deaths_per_million_fixed)

Max total cases, as of now

distinct_loc <- covid %>% distinct(location, .keep_all = T)

total_cases_lookup <- distinct_loc$total_cases
names(total_cases_lookup) <- distinct_loc$location

max_total_cases_lookup <- sapply(distinct_loc$location, function (loc) {
  this_loc <- subset(covid, location == loc)
  return(max(0, this_loc$total_cases, na.rm = T))
})

names(max_total_cases_lookup) <- distinct_loc$location

total_cases_per_million_lookup <- distinct_loc$total_cases_per_million
names(total_cases_per_million_lookup) <- distinct_loc$location

max_total_cases_per_million_lookup <- sapply(distinct_loc$location, function (loc) {
  this_loc <- subset(covid, location == loc)
  return(max(0, this_loc$total_cases_per_million, na.rm = T))
})

names(max_total_cases_per_million_lookup) <- distinct_loc$location

total_deaths_per_million_lookup <- distinct_loc$total_deaths_per_million
names(total_deaths_per_million_lookup) <- distinct_loc$location

max_total_deaths_per_million_lookup <- sapply(distinct_loc$location, function (loc) {
  this_loc <- subset(covid, location == loc)
  return(max(0, this_loc$total_deaths_per_million, na.rm = T))
})

names(max_total_deaths_per_million_lookup) <- distinct_loc$location

max_total_deaths_per_million_lookup['Thailand']

Thailand 
   0.831 

Day over 100

date_over_100_lookup <- sapply(distinct_loc$location, function (loc) {
  this_loc <- subset(covid, location == loc & total_cases_fixed >= 100)
  if (length(this_loc$date) > 0) {
    return(min(this_loc$date))
  }
  
  return(NA)
})

covid$day_over_100 <- sapply(1:nrow(covid), function (n) {
  d <- as.Date(covid$date[n]) - date_over_100_lookup[[covid$location[n]]]
  if (is.na(d) | d < 0) {
    return(NA)
  }
  
  return(d)
})

select(subset(covid, !is.na(day_over_100)), location, date, day_over_100) %>% distinct(location, .keep_all = T)

Top N’s

top_ranks <- data.frame(
  name = names(max_total_cases_per_million_lookup),
  max_total_cases = max_total_cases_lookup,
  max_total_cases_per_million = max_total_cases_per_million_lookup,
  max_total_deaths_per_million = max_total_deaths_per_million_lookup,
  nth_cases = length(max_total_cases_per_million_lookup) - rank(max_total_cases_per_million_lookup) + 1,
  nth_deaths = length(max_total_deaths_per_million_lookup) - rank(max_total_deaths_per_million_lookup) + 1
) %>% arrange(desc(max_total_cases_per_million))

top_ranks %>% select(max_total_cases_per_million)

top_ranks %>% arrange(desc(max_total_deaths_per_million)) %>% select(max_total_deaths_per_million)

Thailand ranks

subset(
  top_ranks %>% arrange(desc(max_total_cases)),
  name %in% (top_ranks %>% arrange(desc(max_total_cases)))$name[2:10] | name == "Thailand"
) %>% select(nth_cases, nth_deaths, max_total_cases_per_million)

Top N, and Thailand, since 100th day

ggplotly(ggplot(
  covid %>% subset(!is.na(day_over_100) & new_cases > 0 & (location %in% top_ranks[1:10,]$name | location == "Thailand")),
  aes(day_over_100, new_cases_per_million, color = location)) 
  + geom_line() 
  + scale_y_log10()
  + labs(
    title = "New cases per million since day over 100 (highest total cases per miliion)",
    x = "Day over 100", y = "Number of new cases per million")
)

ggplotly(ggplot(
  covid %>% subset(!is.na(day_over_100) & new_cases > 0 & (location %in% top_ranks[1:10,]$name | location == "Thailand")),
  aes(day_over_100, new_deaths_per_million, color = location)) 
  + geom_line() 
  + scale_y_log10()
  + labs(
    title = "New deaths per million since day over 100 (highest total cases per miliion)",
    x = "Day over 100", y = "Number of new cases per million")
)

Transformation introduced infinite values in continuous y-axis